Fuzzy Rough Modeling Approach: Based on Fuzzy Clustering and GA Search Strategy

A method to construct fuzzy rough model is proposed. By means of adaptive Gaustafason-Kessel (G-K) clustering algorithm, fuzzy partition can be accomplished and corresponding fuzzy clusters are achieved in data space. Then based on the search of cluster number and attribute subsets through GA search strategy, optimal FRM will be found, and a decision model can be built. The experiment results indicate that FRM method is superior to traditional Bayesian and learning vector quantization (LVQ) methods, moreover, it has more powerful generalization ability. Also, experiment results show that it is favorable to obtain better FRM model if the search of reductive attribute subsets is considered.